codebase-cleanup-deps-audit
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes project manifest files such as package.json and requirements.txt to inventory dependencies, creating a surface where malicious instructions could be embedded in package metadata to influence the agent during analysis.
- Ingestion points: Dependency manifest files (package.json, requirements.txt, Pipfile, go.mod, etc.) identified in resources/implementation-playbook.md.
- Boundary markers: The skill does not explicitly define delimiters to separate ingested data from agent instructions.
- Capability inventory: Network access via the requests library, shell command execution (npm, pip), and file system access.
- Sanitization: Data is parsed using structured formats (JSON, YAML, TOML) but the logic lacks explicit sanitization of field content before further processing.
- [EXTERNAL_DOWNLOADS]: The skill performs network requests to fetch vulnerability data and package metadata from well-known, trusted registries.
- Evidence: Network calls are made to registry.npmjs.org, pypi.org, rubygems.org, ossindex.sonatype.org, and bundlephobia.com in resources/implementation-playbook.md.
- [COMMAND_EXECUTION]: The skill generates and executes shell commands for dependency auditing and automated remediation.
- Evidence: Shell scripts in resources/implementation-playbook.md execute commands including npm audit, npm update, pip install, and pip-compile.
Audit Metadata